Pull Requests ggerganov/llama.cpp

log: add condition of debug log

#24054 opened 2026-06-03 02:10 by ninedreams

cmake: skip cvector-generator and export-lora when CPU backend is disabled model script testing examples python server ggml OpenCL Hexagon

#24053 opened 2026-06-02 21:46 by arichiardi

common: centralise datetime formatting examples server jinja parser

#24047 opened 2026-06-02 20:23 by socram8888

[ggml-webgpu] Implement 2D workgroups for scale, binary, and unary ops ggml WebGPU

#24044 opened 2026-06-02 18:56 by nikhilJain17

server: avoid unnecessary checkpoint invalidation for recurrent / hybrid models examples server

#24035 opened 2026-06-02 16:52 by Regrad

jinja: implement map('filter') testing jinja parser

#24033 opened 2026-06-02 16:08 by MuoDoo

ui: Mermaid Diagrams in chat + interactive preview examples server/ui

#24032 opened 2026-06-02 16:02 by allozaur

tests : add support for qwen3 SSM archs model

#24031 opened 2026-06-02 16:00 by ggerganov

Avoid PDL race conditions by disabling __restrict__ when PDL is used Nvidia GPU ggml

#24030 opened 2026-06-02 15:42 by aendk

server: persist slot checkpoints to .ckpt sidecar (LSCKPT2) examples server

#24028 opened 2026-06-02 15:08 by LuminaNAO

qwen35: use post-norm hidden state for MTP model

#24025 opened 2026-06-02 13:49 by am17an

tools/ui: add OAuth support for MCP servers examples server/ui

#24023 opened 2026-06-02 13:06 by LPFchan

metal : per-op source split + parallel compile ggml Apple Metal

#24021 opened 2026-06-02 12:38 by forforever73

mtmd: correct gemma4 min/max tokens examples

#24014 opened 2026-06-02 09:08 by ngxson

ggml: support concat for scalar types at cuda backend testing Nvidia GPU ggml

#24011 opened 2026-06-02 07:38 by zihaomu

server: add KV cache metrics examples python server

#24010 opened 2026-06-02 06:41 by lvsijian8

[ggml-webgpu] Handle buffer overlap / buffer aliasing for concat operator ggml WebGPU

#24000 opened 2026-06-02 01:40 by nikhilJain17

ggml: add f16 out_prod support for CPU and out_prod op for Vulkan Vulkan ggml

#23997 opened 2026-06-01 21:55 by Lamothe

Make GGML_SYCL_F16=ON the default documentation examples devops ggml SYCL

#23996 opened 2026-06-01 21:23 by malsbat

server/common: Fix `response_format: json_schema` & prefill parsing bug examples server

#23993 opened 2026-06-01 20:04 by roj234

metal: optimize pad_reflect_1d_f32 kernel testing ggml Apple Metal

#23992 opened 2026-06-01 19:58 by shrivasshankar

vulkan: Use cm2 decode_vector for mul_mat_id B matrix loads testing Vulkan ggml

#23991 opened 2026-06-01 19:57 by jeffbolznv

HIP: fixing SSM_SCAN backend testcase error Nvidia GPU ggml

#23983 opened 2026-06-01 16:50 by jiachengjason

server: (router) add model management API examples python server

#23976 opened 2026-06-01 14:45 by ngxson

build : use umbrella Headers directory for XCFramework module map

#23974 opened 2026-06-01 14:03 by gmarzjr

vulkan: add fast path for contiguous buffer transfers Vulkan ggml

#23973 opened 2026-06-01 13:47 by winstonma

vulkan: add fwht support for Intel with shmem reduction Vulkan ggml

#23964 opened 2026-06-01 10:33 by 0cc4m

ggml-cpu : add AVX2 and AVX optimization for nvfp4 dot product ggml

#23961 opened 2026-06-01 08:24 by ragz4125

common: vectorize common_embd_similarity_cos for powerpc.

#23960 opened 2026-06-01 07:40 by shalinib-ibm

Simple implementation for limiting shell commands. examples server

#23956 opened 2026-06-01 05:00 by Penguin-Guru